Make decoding reusable for LLDB and GDB

Author: mborland

extra/decimal_printer.py

@@ -1,5 +1,8 @@
-# decimal_printer.py
+# Copyright 2025 Matt Borland
+# Distributed under the Boost Software License, Version 1.0.
+# https://www.boost.org/LICENSE_1_0.txt
 
+from detail.decode_decimal32 import decode_decimal32
 import lldb
 
 def decimal32_summary(valobj, internal_dict):
@@ -11,56 +14,8 @@ def decimal32_summary(valobj, internal_dict):
     try:
         val = valobj.GetNonSyntheticValue()
         bits = val.GetChildMemberWithName("bits_").GetValueAsUnsigned()
+        return decode_decimal32(bits)
 
-        sign = bits & 2147483648 != 0
-        isnan = False
-
-        if bits & 2013265920 == 2013265920:
-
-            if bits & 2113929216 == 2113929216:
-                result = "-SNAN" if sign else "SNAN"
-                isnan = True
-            elif bits & 2080374784 == 2080374784:
-                result = "-SNAN" if sign else "QNAN"
-                isnan = True
-            elif bits & 2080374784 == 2013265920:
-                result = "-INF" if sign else "INF"
-            else:
-                raise ValueError("Unknown Finite Value")
-
-            if isnan:
-                payload = bits & 8388607
-                if payload > 0:
-                    result += '(' + str(payload) + ')'
-
-        else:
-            # See decimal32_t::to_components()
-            d32_comb_11_mask = 1610612736
-            if bits & d32_comb_11_mask == d32_comb_11_mask:
-                implied_bit = 8388608
-                significand = implied_bit | (bits & 2097151)
-                exp = (bits & 534773760) >> 21
-            else:
-                significand = bits & 8388607
-                exp = (bits & 2139095040) >> 23
-
-            exp -= 101 # Bias Value
-
-            if significand == 0:
-                result = "0.0e+0"
-            else:
-                n_digits = len(str(abs(significand)))
-
-                # If there is no fractional component we want to remove the decimal point and zero
-                if n_digits > 1:
-                    normalized = significand / (10 ** (n_digits - 1))
-                    total_exp = exp + n_digits - 1
-                else:
-                    normalized = significand
-                    total_exp = exp
-                result = f"{'-' if sign else ''}{normalized:.{n_digits - 1}f}e{total_exp:+}"
-
-        return result
     except Exception as e:
         return f"<invalid decimal32_t: {e}>"
 

extra/detail/decode_decimal32.py

@@ -0,0 +1,54 @@
+# Copyright 2025 Matt Borland
+# Distributed under the Boost Software License, Version 1.0.
+# https://www.boost.org/LICENSE_1_0.txt
+
+def decode_decimal32(bits):
+    sign = bits & 2147483648 != 0
+    isnan = False
+
+    if bits & 2013265920 == 2013265920:
+
+        if bits & 2113929216 == 2113929216:
+            result = "-SNAN" if sign else "SNAN"
+            isnan = True
+        elif bits & 2080374784 == 2080374784:
+            result = "-SNAN" if sign else "QNAN"
+            isnan = True
+        elif bits & 2080374784 == 2013265920:
+            result = "-INF" if sign else "INF"
+        else:
+            raise ValueError("Unknown Finite Value")
+
+        if isnan:
+            payload = bits & 8388607
+            if payload > 0:
+                result += '(' + str(payload) + ')'
+
+    else:
+        # See decimal32_t::to_components()
+        d32_comb_11_mask = 1610612736
+        if bits & d32_comb_11_mask == d32_comb_11_mask:
+            implied_bit = 8388608
+            significand = implied_bit | (bits & 2097151)
+            exp = (bits & 534773760) >> 21
+        else:
+            significand = bits & 8388607
+            exp = (bits & 2139095040) >> 23
+
+        exp -= 101 # Bias Value
+
+        if significand == 0:
+            result = "0.0e+0"
+        else:
+            n_digits = len(str(abs(significand)))
+
+            # If there is no fractional component we want to remove the decimal point and zero
+            if n_digits > 1:
+                normalized = significand / (10 ** (n_digits - 1))
+                total_exp = exp + n_digits - 1
+            else:
+                normalized = significand
+                total_exp = exp
+            result = f"{'-' if sign else ''}{normalized:.{n_digits - 1}f}e{total_exp:+}"
+
+    return result